Kouhei Sutou
null+****@clear*****
Thu Apr 5 18:21:31 JST 2018
Kouhei Sutou 2018-04-05 18:21:31 +0900 (Thu, 05 Apr 2018) New Revision: edf9b24b80b1caa4db3f93852d7aed3966540d92 https://github.com/groonga/groonga/commit/edf9b24b80b1caa4db3f93852d7aed3966540d92 Message: schema: support dumping tokenizer with options Added files: test/command/suite/schema/tables/tokenizer_with_options.test Copied files: test/command/suite/schema/tables/tokenizer_with_options.expected (from test/command/suite/schema/tables/tokenizer.expected) Modified files: lib/proc/proc_schema.c test/command/suite/schema/tables/columns/type/index_medium.expected test/command/suite/schema/tables/columns/type/index_small.expected test/command/suite/schema/tables/token_filters.expected test/command/suite/schema/tables/tokenizer.expected Modified: lib/proc/proc_schema.c (+83 -7) =================================================================== --- lib/proc/proc_schema.c 2018-04-05 18:18:29 +0900 (eb1e71943) +++ lib/proc/proc_schema.c 2018-04-05 18:21:31 +0900 (a9b527c1e) @@ -1,6 +1,6 @@ /* -*- c-basic-offset: 2 -*- */ /* - Copyright(C) 2015-2016 Brazil + Copyright(C) 2015-2018 Brazil This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public @@ -97,6 +97,30 @@ command_schema_output_value_type(grn_ctx *ctx, grn_obj *value_type) command_schema_output_type(ctx, "value_type", value_type); } +static grn_bool +command_schema_output_command_include_special_character(grn_ctx *ctx, + const char *value, + unsigned int size) +{ + const char *end = value + size; + + for (; value < end; value++) { + switch (value[0]) { + case '(' : + case ')' : + case ' ' : + case '"' : + case '\'' : + return GRN_TRUE; + break; + default : + break; + } + } + + return GRN_FALSE; +} + static void command_schema_output_command(grn_ctx *ctx, const char *command_name, @@ -145,12 +169,22 @@ command_schema_output_command(grn_ctx *ctx, name_size = grn_vector_get_element(ctx, arguments, i, &name, NULL, NULL); + grn_text_printf(ctx, &command_line, + " --%.*s ", + name_size, name); value_size = grn_vector_get_element(ctx, arguments, i + 1, &value, NULL, NULL); - grn_text_printf(ctx, &command_line, - " --%.*s %.*s", - name_size, name, - value_size, value); + if (command_schema_output_command_include_special_character(ctx, + value, + value_size)) { + grn_obj value_text; + GRN_TEXT_INIT(&value_text, GRN_OBJ_DO_SHALLOW_COPY); + GRN_TEXT_SET(ctx, &value_text, value, value_size); + grn_text_otoj(ctx, &command_line, &value_text, NULL); + GRN_OBJ_FIN(ctx, &value_text); + } else { + GRN_TEXT_PUT(ctx, &command_line, value, value_size); + } } grn_ctx_output_str(ctx, GRN_TEXT_VALUE(&command_line), @@ -506,7 +540,7 @@ command_schema_table_output_tokenizer(grn_ctx *ctx, grn_obj *table) return; } - grn_ctx_output_map_open(ctx, "tokenizer", 2); + grn_ctx_output_map_open(ctx, "tokenizer", 3); grn_ctx_output_cstr(ctx, "id"); command_schema_output_id(ctx, tokenizer); @@ -514,6 +548,43 @@ command_schema_table_output_tokenizer(grn_ctx *ctx, grn_obj *table) grn_ctx_output_cstr(ctx, "name"); command_schema_output_name(ctx, tokenizer); + grn_ctx_output_cstr(ctx, "options"); + { + grn_obj options; + unsigned int n; + + GRN_VOID_INIT(&options); + grn_table_get_tokenizer_options(ctx, table, &options); + if (options.header.type == GRN_VOID) { + grn_ctx_output_null(ctx); + } else { + grn_obj option; + unsigned int i; + + n = grn_vector_size(ctx, &options); + grn_ctx_output_array_open(ctx, "options", n); + GRN_VOID_INIT(&option); + for (i = 0; i < n; i++) { + const char *value; + unsigned int length; + grn_id domain; + + length = grn_vector_get_element(ctx, + &options, + i, + &value, + NULL, + &domain); + grn_obj_reinit(ctx, &option, domain, 0); + grn_bulk_write(ctx, &option, value, length); + grn_ctx_output_obj(ctx, &option, NULL); + } + GRN_OBJ_FIN(ctx, &option); + grn_ctx_output_array_close(ctx); + } + GRN_OBJ_FIN(ctx, &options); + } + grn_ctx_output_map_close(ctx); } @@ -642,7 +713,12 @@ command_schema_table_command_collect_arguments(grn_ctx *ctx, grn_obj *tokenizer; tokenizer = grn_obj_get_info(ctx, table, GRN_INFO_DEFAULT_TOKENIZER, NULL); if (tokenizer) { - ADD_OBJECT_NAME("default_tokenizer", tokenizer); + grn_obj sub_output; + GRN_TEXT_INIT(&sub_output, 0); + grn_table_get_tokenizer_string(ctx, table, &sub_output); + GRN_TEXT_PUTC(ctx, &sub_output, '\0'); + ADD("default_tokenizer", GRN_TEXT_VALUE(&sub_output)); + GRN_OBJ_FIN(ctx, &sub_output); } } Modified: test/command/suite/schema/tables/columns/type/index_medium.expected (+2 -1) =================================================================== --- test/command/suite/schema/tables/columns/type/index_medium.expected 2018-04-05 18:18:29 +0900 (246983f77) +++ test/command/suite/schema/tables/columns/type/index_medium.expected 2018-04-05 18:21:31 +0900 (f028fb7cb) @@ -338,7 +338,8 @@ schema "value_type": null, "tokenizer": { "id": 67, - "name": "TokenBigram" + "name": "TokenBigram", + "options": null }, "normalizer": { "id": 79, Modified: test/command/suite/schema/tables/columns/type/index_small.expected (+2 -1) =================================================================== --- test/command/suite/schema/tables/columns/type/index_small.expected 2018-04-05 18:18:29 +0900 (d1d7bb133) +++ test/command/suite/schema/tables/columns/type/index_small.expected 2018-04-05 18:21:31 +0900 (0326d2757) @@ -338,7 +338,8 @@ schema "value_type": null, "tokenizer": { "id": 67, - "name": "TokenBigram" + "name": "TokenBigram", + "options": null }, "normalizer": { "id": 79, Modified: test/command/suite/schema/tables/token_filters.expected (+2 -1) =================================================================== --- test/command/suite/schema/tables/token_filters.expected 2018-04-05 18:18:29 +0900 (82f07365b) +++ test/command/suite/schema/tables/token_filters.expected 2018-04-05 18:21:31 +0900 (620433ea3) @@ -227,7 +227,8 @@ schema "value_type": null, "tokenizer": { "id": 67, - "name": "TokenBigram" + "name": "TokenBigram", + "options": null }, "normalizer": null, "token_filters": [ Modified: test/command/suite/schema/tables/tokenizer.expected (+2 -1) =================================================================== --- test/command/suite/schema/tables/tokenizer.expected 2018-04-05 18:18:29 +0900 (55128d980) +++ test/command/suite/schema/tables/tokenizer.expected 2018-04-05 18:21:31 +0900 (b7cfcd2d2) @@ -218,7 +218,8 @@ schema "value_type": null, "tokenizer": { "id": 67, - "name": "TokenBigram" + "name": "TokenBigram", + "options": null }, "normalizer": null, "token_filters": [ Copied: test/command/suite/schema/tables/tokenizer_with_options.expected (+9 -5) 93% =================================================================== --- test/command/suite/schema/tables/tokenizer.expected 2018-04-05 18:18:29 +0900 (55128d980) +++ test/command/suite/schema/tables/tokenizer_with_options.expected 2018-04-05 18:21:31 +0900 (41ed03ce6) @@ -1,4 +1,4 @@ -table_create Terms TABLE_PAT_KEY ShortText --default_tokenizer TokenBigram +table_create Terms TABLE_PAT_KEY ShortText --default_tokenizer 'TokenNgram("n", 4)' [[0,0.0,0.0],true] schema [ @@ -217,8 +217,12 @@ schema }, "value_type": null, "tokenizer": { - "id": 67, - "name": "TokenBigram" + "id": 78, + "name": "TokenNgram", + "options": [ + "n", + 4 + ] }, "normalizer": null, "token_filters": [ @@ -233,9 +237,9 @@ schema "name": "Terms", "flags": "TABLE_PAT_KEY", "key_type": "ShortText", - "default_tokenizer": "TokenBigram" + "default_tokenizer": "TokenNgram(\"n\", 4)" }, - "command_line": "table_create --name Terms --flags TABLE_PAT_KEY --key_type ShortText --default_tokenizer TokenBigram" + "command_line": "table_create --name Terms --flags TABLE_PAT_KEY --key_type ShortText --default_tokenizer \"TokenNgram(\\\"n\\\", 4)\"" }, "columns": { } Added: test/command/suite/schema/tables/tokenizer_with_options.test (+4 -0) 100644 =================================================================== --- /dev/null +++ test/command/suite/schema/tables/tokenizer_with_options.test 2018-04-05 18:21:31 +0900 (29d176baa) @@ -0,0 +1,4 @@ +table_create Terms TABLE_PAT_KEY ShortText \ + --default_tokenizer 'TokenNgram("n", 4)' + +schema -------------- next part -------------- HTML����������������������������... URL: https://lists.osdn.me/mailman/archives/groonga-commit/attachments/20180405/251dbc30/attachment-0001.htm
